Detecting economically important palms using UAV imagery in intact, moist tropical forest

Palm trees are important resources in the moist tropical forest due to the provisioning ecosystem services that they supply, especially for fruit production. Some of these fruits are considered as “super foods” due to their rich nutritional values, and they are an important food for both local communities and fauna. A common constraint to expanding sustainable management of palms in intact forest has been the difficulty of mapping their abundance and distribution at large scales. Typical ground-based surveys sample small areas, while management decisions require precise information at larger scales. In recent years, small Unmanned Aerial Vehicles (UAVs) have become an important tool for mapping forest areas as they are cheap and easy to transport, and they provide high spatial resolution imagery of remote and difficult-to-access areas. This study combined field data and RGB UAV imagery to identify and delineate palm tree crowns in intact forest in the Peruvian Amazon. Nine permanent RAINFOR plots with 1,472 reference palm trees were flown with a Phantom 4Pro UAV from October to December 2017 in the Loreto Region, Peru. The results indicate that the textural information obtained from the RGB imagery combined with the canopy height model can identify important palm species like Mauritia flexuosa, Euterpe precatoria and Oenocarpus bataua with an overall accuracy of 86% using a support vector machine radial algorithm. However, since the UAV camera only takes pictures of the canopy, on average, only 80% of the referenced palm trees were identified, and understorey palms were often missed. The integration of field and UAV data has the potential of providing precise estimates of resource availability at scales relevant to forest management, especially where cloud cover limits the use of satellite imagery, and the large areas and accessibility restrict ground-based surveys.